9
$\begingroup$

We all known the problems that presents the notation of inverse/reverse/anti functions as $f^{-1}(x)$, being the most important one the confusion with ${f(x)}^{-1}$, as in the classical $\sin^{-1}(x)$, or the strange cases such as ${f^{-1}(x)}^{-2}$,etc. Problems appears strongly in manual writing.

There are any other notation convention for the inverse functions more or less widely used ? Something like $f^{\#}(x)$, $f^{r}(x)$, $rev\{f\}(x)$?

If not, why not ? It seems something interesting to improve.

$\endgroup$
  • 3
    $\begingroup$ There is nothing standard along those lines. There might be some advantage in having a notation like the ones you propose but I think it would be difficult to get one adopted $\endgroup$ – Ethan Bolker Feb 23 '18 at 12:57
  • $\begingroup$ Suppose that there are $n$ ways to write the inverse function (yours not included). Then if one of your proposed notations will be accepted there will be $n+1$ ways to write the inverse function. Not such a good idea, is it? Don't think that it is achievable to ban the other notations out. $\endgroup$ – drhab Feb 23 '18 at 13:01
  • 1
    $\begingroup$ Are you're probably aware, there are common special notations for inverses of certain special functions, e.g., $\arctan$, $\operatorname{arsinh}$, etc. (Of course, $\arctan$ is not the inverse of $\tan$ but rather its restriction to a certain interval, making the notation $\tan^{-1}$ even more troublesome; for this reason I strongly prefer $\arctan$, especially for teaching.) $\endgroup$ – Travis Feb 23 '18 at 13:03
  • 1
    $\begingroup$ Perhaps something that hints to reflection across the function $g(x)=x$ would be nice. $\endgroup$ – Samuel Feb 23 '18 at 13:06
  • 2
    $\begingroup$ The real problem is that $\sin^2 x \neq \sin \sin x$ $\endgroup$ – Vaelus Feb 23 '18 at 17:42
14
$\begingroup$

If anything, I think $f^{-1}(x)$ is absolutely the correct notation for an inverse function. Correspondingly, I think $f^2(x)$ is absolutely the correct notation for $(f\circ f)(x) = f(f(x))$, not for $(f(x))^2$. But this is definitely a matter of taste, as well as context, and other people will disagree with me. For example, if $f$ isn't an endomorphism then $(f\circ f)$ makes no sense, whereas if the codomain of $f$ doesn't have any kind of multiplicative structure then $(f(x))^2$ doesn't make sense. As a reader, you should always keep in mind that the author may use symbols in a way more appropriate to their work (or even just their aesthetic sensibilities) and be willing to do some interpretative work!

There might be some sense in defining e.g. $f^{\circ 2}(x) = (f\circ f)(x)$, $f^{\cdot 2}(x) = (f(x))^2$, similarly to how we define $f^{\otimes 2}(x) = (f\otimes f)(x)$ on a tensor square of vector spaces $V^{\otimes 2}$, etc. I've seen this notation before. This would make the inverse function $f^{\circ(-1)}(x)$ and the reciprocal $f^{\cdot(-1)}(x)$. But this is pretty ugly, and I've almost never seen it. Unless you have a whole bunch of binary operations like $\circ, \cdot, \otimes$ lying around and you need to explicitly disambiguate between them, I'd avoid it. Just use words. (Besides which, you'd probably need to use words to tell readers what your notation means, because it's probably new to them.)

Good notation is like gold dust, but keep in mind that most notation is bad. Don't add to it, if possible. Just make your own work as readable as you can.

$\endgroup$
  • $\begingroup$ All received answers are interesting. I will choice this one as answer (to do not leave behind unanswered question) because it proposes a feasible alternative. I will try, in my particular writings, use an exponent number inside a circle for composition, by example $f^{②}(x)$. As an optional short cut for reverse "-1" the notation $f^{\circ}(x)$. $\endgroup$ – pasaba por aqui Feb 24 '18 at 10:35
8
$\begingroup$

The reason for the two conventional uses of $^{-1}$ is that what they invert are written very similarily. In fact, in many settings, composition of two functions is actually written exactly like multiplication.

For instance, multiplying matrices corresponds to composing linear maps, so we often see composed linear maps written next to one another without a symbol between, as though we're multiplying. And when working with symmetry groups, a composition of two permutations is also often written by writing them next to one another without a symbol in-between.

So, seing as how composition of functions is often written like multiplication, it is not that difficult to see why the compositional inverse is also written the same way as the multiplicative inverse.

I would love for there to be a consistent way to differentiate between the two, but there isn't. At least not any widely used convention. Also, in some cases, like in my two examples above, it would mean you have to decide, every time you want to invert an object, whether that object is a function to be composed, or an element of a group or monoid to be multiplied, when really, you want it to be both at the same time.

$\endgroup$
7
$\begingroup$

I was reading Tensor Geometry not too long ago and I encountered a good notation for the reverse of a map,

enter image description here

$\endgroup$
  • 4
    $\begingroup$ It's clear from context that you're aware of this, but for OP's benefit: As defined $f^{\leftarrow}$ is not the inverse of $f$, but is a map $\mathcal P(Y) \to \mathcal P(X)$ (here $\mathcal P(Z)$ is the set of all subsets of $Z$). Of course, if $f$ is bijective, then the inverse of $f$ is $\pi \circ f^{\leftarrow}$, where $\pi$ is the map $\{\{x\} : x \in X\} \to X$ that maps a singleton set to its sole element. That said, $f^{-1}$ is already used for both $f^{\leftarrow}$ and $\pi\circ f^{\leftarrow}$, so there's no reason we couldn't abuse the notation and use $f^{\leftarrow}$ for both too. $\endgroup$ – Travis Feb 23 '18 at 13:57
  • 1
    $\begingroup$ @Travis: I know, I know. I was going to add an except about how sometimes abusing notation is not really abuse, but I wanted to keep things short. Thanks for the comment. Hopefully nothing I said is misleading to the OP. $\endgroup$ – Faraad Armwood Feb 23 '18 at 14:03
6
$\begingroup$

Suppose we have previously established, $f:U \rightarrow V$, then unambiguous notation for the inverse can be introduced by something like "Let $g:V \rightarrow U$ be the inverse of $f$."

Standard notations already compress information extensively, sometimes to the edge of comprehensibility. Attempting further compression isn't a clear improvement to communication.

$\endgroup$
  • $\begingroup$ I'd introduce my own syntax with description such as: given $f:U \rightarrow V$ let $\hat{f}: V \rightarrow U$ the inverse of $f$. $\endgroup$ – Jakob Mar 3 at 21:04
4
$\begingroup$

It is a cute suggestion. However, the idea of the current notation comes from a convenient shorthand $$f(f(x)) = f^2(x)$$ which allow easy "exponent" arithmetic like $$f^{-1}(f(f(x)) = f^{2-1}(x) = f(x).$$

So the current notation also has some useful features, which is the reason it was invented in the first place.

Personally, I try avoid ambiguity by distinguishing $f^{-1}(x)$ denoting the functional inverse from $f(x)^{-1}$ denoting the multiplicative inverse. In addition, for the trig functions you indicated, I use $\arcsin(x)$ instead of $\sin^{-1}(x)$, which has an additional usefulness in complex analysis since you can indicate the principal branch limitation by $\mathrm{Arcsin}(x)$, thereby distinguishing it from the general case...

$\endgroup$
  • 2
    $\begingroup$ Thanks for your answer. $cos^2 \alpha+ sin^2 \alpha = 1$ is an usual expression that breaks all rules. Ok, as you said, if we remember that arcxxx exist and that cos(cos(x)) is near than never used, we can accept a notation for trigonometry and another for "remainder" math areas. But it is a few confusing, in special, in high school. $\endgroup$ – pasaba por aqui Feb 23 '18 at 22:12
  • $\begingroup$ @pasabaporaqui agreed, may be confusing in the beginning, but eventually the context implies reasonable convention... $\endgroup$ – gt6989b Feb 25 '18 at 19:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.